Article Cite This: Environ. Sci. Technol. 2019, 53, 8027−8035
pubs.acs.org/est
Hydrocarbons in Upland Groundwater, Marcellus Shale Region, Northeastern Pennsylvania and Southern New York, U.S.A. Peter B. McMahon,*,† Bruce D. Lindsey,‡ Matthew D. Conlon,‡ Andrew G. Hunt,† Kenneth Belitz,§ Bryant C. Jurgens,∥ and Brian A. Varela† †
U.S. U.S. § U.S. ∥ U.S.
Downloaded via UNIV OF SOUTHERN INDIANA on July 17, 2019 at 09:01:49 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
Geological Geological Geological Geological
Survey, Survey, Survey, Survey,
Lakewood, Colorado 80225, United States New Cumberland, Pennsylvania 17070, United States Northboro, Massachusetts 01532, United States Sacramento, California 95819, United States
S Supporting Information *
ABSTRACT: Water samples from 50 domestic wells located 1 km (distal) from shale-gas wells in upland areas of the Marcellus Shale region were analyzed for chemical, isotopic, and groundwater-age tracers. Uplands were targeted because natural mixing with brine and hydrocarbons from deep formations is less common in those areas compared to valleys. CH4-isotope, predrill CH4-concentration, and other data indicate that one proximal sample (5% of proximal samples) contains thermogenic CH4 (2.6 mg/L) from a relatively shallow source (Catskill/Lock Haven Formations) that appears to have been mobilized by shale-gas production activities. Another proximal sample contains five other volatile hydrocarbons (0.03−0.4 μg/L), including benzene, more hydrocarbons than in any other sample. Modeled groundwater-age distributions, calibrated to 3H, SF6, and 14 C concentrations, indicate that water in that sample recharged prior to shale-gas development, suggesting that land-surface releases associated with shale-gas production were not the source of those hydrocarbons, although subsurface leakage from a nearby gas well directly into the groundwater cannot be ruled out. Age distributions in the samples span ∼20 to >10000 years and have implications for relating occurrences of hydrocarbons in groundwater to land-surface releases associated with recent shale-gas production and for the time required to flush contaminants from the system.
■
INTRODUCTION Understanding the effects of unconventional oil and gas (UOG) production on groundwater quality is an important environmental issue in the United States (U.S.) and other countries where UOG resources are present.1−3 The Middle Devonian Marcellus Shale, in the northern Appalachian Basin (NAB) in the northeastern U.S., has been the target of substantial UOG development since about 2005.4,5 Although the quality of shallow groundwater overlying the Marcellus Shale has been extensively studied, debate continues as to whether UOG activities, other types of anthropogenic activity, or natural processes are primarily responsible for the occurrence of hydrocarbons in some shallow groundwater.4,6−9 Several studies in the Marcellus region showed that elevated concentrations of thermogenic methane (CH4) occur in shallow groundwater in river valleys and near some faults and other geologic structures, due to natural mixing with gasrich brine in those areas.5,10−14 Other occurrences of thermogenic CH4 in shallow groundwater have been attributed to leakage from compromised gas wells;6,7,15,16 however, definitive source attribution can be challenging in areas where natural and anthropogenic processes potentially overlap. In such areas, baseline data collected prior to gas-well drilling8 © 2019 American Chemical Society
and other comprehensive geochemical data sets supported by detailed information on nearby gas wells have been used to understand occurrences of thermogenic CH4 in groundwater.15,17 While thermogenic CH4 in groundwater is typically from subsurface sources, studies have also detected other organic compounds in shallow groundwater in the region that appear to be from land-surface sources.18,19 A recent study attributed elevated concentrations of diesel-range organic compounds (DROs) in groundwater to accidental spills of hydraulic fracturing fluid at land surface.18 That conclusion was based on multiple lines of evidence, including the fact that the highest DRO concentrations occurred in shallow, tritium (3H)active groundwater (i.e., young groundwater).18 The presence of 3H in groundwater is clear evidence of post-1950 recharge in the aquifer, but 3H occurrence alone is not necessarily indicative of very young water due to the history of declining atmospheric 3H inputs, 3H decay, and groundwater mixing processes.20−22 The question of groundwater age is relevant in Received: Revised: Accepted: Published: 8027
March 7, 2019 June 9, 2019 June 19, 2019 June 27, 2019 DOI: 10.1021/acs.est.9b01440 Environ. Sci. Technol. 2019, 53, 8027−8035
Article
Environmental Science & Technology
Figure 1. Concentrations of (A) bromide in relation to chloride, (B) sulfur hexafluoride in relation to tritium, and (C) corrected carbon-14 of dissolved inorganic carbon in relation to tritium. Part A includes data for Types A and D water from reference 11 and data for brines from references 11 and 31−33. Parts B and C include dispersion-model curves from reference 30; dashed lines represent mixing between post-1950 water with a mean age of 25 years and pre-1950 water with mean ages of 2500 and 15000 years.
area.6,7,15 The 2015 IHS Markit data set for oil and gas wells and State databases were used to determine the number and distance of hydrocarbon wells from sampled water wells.24−26 One of the 50 water wells may have a conventional gas well within 1 km. Water wells were sampled after purging three casing volumes of water and when the readings of pH, water temperature, specific conductance, and dissolved O2 were stabilized. As described in SI Section 2, water samples were collected and analyzed for major ions and trace elements; isotopic compositions of water (δ2H-H2O, δ18O-H2O) and dissolved inorganic carbon (δ13C-DIC); VOCs (21 compounds); noble gas abundances (helium-4 (4He), neon (Ne), argon (Ar), krypton (Kr), xenon (Xe)), and isotopes ( 3 He/ 4 He, 20 Ne/22Ne, 40Ar/36Ar, 86Kr/84Kr, 130Xe/132Xe); groundwaterage tracers (3H), sulfur hexafluoride (SF6), carbon-14 in DIC (14C-DIC); concentrations of CH4 through pentane (C1−C5) and the isotopic composition of CH4 (δ2H-CH4, δ13C-CH4). Data are listed in SI Tables S2 and S3 and are available in electronic format in references 27 and 28. Kruskal−Wallis tests, as implemented in the software OriginPro 2018,29 were used on ranked data to test for significant differences in methane concentrations between study areas. Spearman correlation analysis was used to examine relations between concentrations of methane and other variables. For the statistical tests, concentrations below reporting levels were set to zero. An α value of 0.05 was used for each test. Age-tracer concentrations were modeled using the software TracerLPM to determine fractions of post-1950 groundwater in the samples, mean ages of the pre- and post-1950 fractions, and age distributions in the samples.21,23,30 Pre- and post-1950 groundwaters are defined as water recharged before or after the early 1950s start of above ground nuclear weapons testing, respectively. The age-dating analysis is described in SI Section 3, and results are in Table S4.
the Marcellus region where UOG development is relatively recent, and it could be important with respect to overall aquifer vulnerability to contamination from UOG surface sources.21−23 Upland areas in the Marcellus region are well suited to studying the occurrence of hydrocarbons in shallow groundwater from UOG-related subsurface and land-surface sources. Groundwater in the upland areas typically contains much less thermogenic CH4 from natural subsurface sources than groundwater in valleys due to the uplands being recharge areas,10,12,14 potentially making it easier to identify CH4 associated with UOG activities in uplands. Because they are recharge areas, the uplands could be expected to have a high degree of connectivity to the land surface and any associated hydrocarbon sources. In 2017, we sampled 50 domestic wells in upland areas of southern New York (NY), where Marcellus UOG development is currently banned, and northeastern Pennsylvanian (PA), where there is active UOG development (Supporting Information (SI) Figure S1). Groundwater samples were analyzed for a broad suite of chemical and isotopic tracers, hydrocarbon gases, volatile organic compounds (VOCs), noble gases, and groundwater-age tracers. The purpose of this study is to determine if hydrocarbons related to UOG activities are present in upland groundwater. Unlike many previous Marcellus studies, we focus on areas relatively unaffected by natural mixing with hydrocarbonbearing fluids and use more refined groundwater-age estimates to better understand the timing of contamination from potential land-surface sources.
■
MATERIALS AND METHODS Uplands are defined as areas >325 m in elevation and >300 m from rivers (SI Section 1, Figure S1), which generally excludes most valleys from the study area.10 In NY, the domestic wells (n = 15) primarily produce from the Java-West Falls and Sonyea Formations; and in PA, the wells (n = 35) primarily produce from the Catskill and Lock Haven Formations (Table S1). The aquifers are generally composed of fractured Upper Devonian shale, siltstone, and sandstone,4,10,13 forming a dualporosity flow system containing fracture and rock-matrix porosity.10 The 15 NY wells and 15 PA wells are located >1 km from active UOG wells (NY distal−ND and PA distal−PD, respectively). Twenty additional PA wells are located
